Ben (@ben_frigon) header
Ben (@ben_frigon) profile picture
Joined December 2024
The vibe has shifted. The President launched a memecoin. Snoop Dog is singing Journey. Crypto will be made in America. AI will be made in America. Don’t stop believin
141
189
2
830
23,926
Is nuclear power the ONLY solution for continuing to grow AI? Check out what Talus Co-Founder and CSO @ben_frigon said when he recently joined @KevinWSHPod on DROPS 👇
13
26
1
50
4,594
I will be following Yann LeCun’s new company with great interest. Perhaps world models will win.
10
14
1,146
In the long run, longs win
The manager skill debate is largely a matter of datasets. -GS report looking at 5 years (2020-2024) showed traditional LO ($28t in AUM) underperformed. -HF long ($600b in AUM) produced 200-300 bps of outperformance. -130/30 active extension HFs outperformed even more.
5
1
9
832
If you thought 2025 was a great year for AI 2026 will be even better
2
6
514
If you thought 2025 was a great year for AI 2026 will be even better
7
2
1
21
1,951
I've never felt this much behind as a programmer. The profession is being dramatically refactored as the bits contributed by the programmer are increasingly sparse and between. I have a sense that I could be 10X more powerful if I just properly string together what has become available over the last ~year and a failure to claim the boost feels decidedly like skill issue. There's a new programmable layer of abstraction to master (in addition to the usual layers below) involving agents, subagents, their prompts, contexts, memory, modes, permissions, tools, plugins, skills, hooks, MCP, LSP, slash commands, workflows, IDE integrations, and a need to build an all-encompassing mental model for strengths and pitfalls of fundamentally stochastic, fallible, unintelligible and changing entities suddenly intermingled with what used to be good old fashioned engineering. Clearly some powerful alien tool was handed around except it comes with no manual and everyone has to figure out how to hold it and operate it, while the resulting magnitude 9 earthquake is rocking the profession. Roll up your sleeves to not fall behind.
2,554
7,301
2,906
55,184
17,115,700
Brown Classics department is undefeated
Ted Turner (now a billionaire) wanted to major in Classics in college. You won’t believe the letter his father sent him. (1/3)
1
4
392
Is AI a bubble, or are there still big opportunities? 🫧 Check out what Talus Co-Founder and CSO @ben_frigon said when he recently joined @KevinWSHPod on DROPS 👇
20
5
74
5,142
1. Research 2. Strategy 3. Tactics 4. Execution
34
5
1
27
1,626
Whoever did this, thank you!❤️🐻🙏
4
36
2
459
36,347
I'll tell two stories about BH. First, I took ENGN 9 there, taught by the legendary Barrett Hazeltine, in the same classroom the shooting supposedly happened. It is probably the most famous class at Brown and helped spark my interest in entrepreneurship. Many Bruno founders would say the same. Second, there is an apple tree outside BH that they say is a descendant of Isaac Newton’s tree. Yes, the famous apple tree he was sitting under when an apple hit him in the head and made him think of gravity. I took a picture of it the first time I saw it bc I didn't believe it lol. There are many strange figures and monuments on campus but I was always incredulous at the Isaac Newton tree. God bless Brown. It is a very special place. Praying for the community to come together and heal. #EverTrue
I spent many hours inside Barus and Holley. It is the engineering building and it was one of my favorite places on campus. I have some great memories of that place. I'm so sad to know that students just living their lives like I did had something so terrible happen to them.
2
5
941
I spent many hours inside Barus and Holley. It is the engineering building and it was one of my favorite places on campus. I have some great memories of that place. I'm so sad to know that students just living their lives like I did had something so terrible happen to them.
Brown University student Joseph Oduro was leading a review session for an upcoming economics final with around 60 other students when he says a masked man entered the lecture hall armed with a large gun. At least two people were killed and nine others were injured in Saturday's shooting, according to officials. "I was standing in the front so as soon as he walked in," Oduro said. "He immediately saw me and I immediately saw him." Read more: abcnews.link/Wd8ayXF
1
2
8
2,276
I'm heartbroken by this terrible tragedy at Brown. Praying for the victims, their families, and the school.
WATCH: Emotional crowd gathers for a candlelight vigil the day after a deadly mass shooting at Brown University.
6
514
Patriots hold moment of silence for Brown University shooting victims trib.al/opseSPV
7
9
78
17,991
Blame the EVM Then start building with @SuiNetwork Move
this is so f*cking scary - a txn made 9 months ago to transfer just $5 ended up causing a $30K loss. TLDR: → a txn was approved 9 months ago → to interact with the ThirdWeb Universal Bridge → Infinite approval was given (just for a $5 transfer) → a vulnerability in the contract was discovered in April → but the contract remained active onchain, including the blog referencing the affected contract address → vulnerability allowed anyone to drain funds from users with infinite approval → the hacker waited 9 months for the user to deposit a substantial amount → And boom - $30K gone in a single txn → funds were sent to Railgun, so can’t be traced → thirdWeb permanently disabled the legacy affected contract after the hack was disclosed → the blog referencing the affected (now-disabled) contract address is still live I don’t even know whom to blame here. It’s honestly a shame that after all these years of building, crypto users still have to deal with stuff like this and this user is literally one of the smartest people in the industry.
2
10
520
Why AI Needs Trust, Nuclear Power, and Crypto-Native Agents In this episode of DROPS, I sit down with Ben Frigon to discuss why AI might soon automate huge parts of your life, but only if we solve three hard problems at once: trust, energy, and infrastructure. From nuclear power and data ownership to AI agents that move money while you sleep, @ben_frigon sketches a future that feels ambitious yet oddly practical. From DeFi Summer to @TalusNetwork Ben’s story starts at Brown University during DeFi summer, when Ethereum and early DeFi apps were exploding. He became president of the blockchain club and co-designed what he calls “the first credit-bearing course on NFTs at any college anywhere,” taught right at the top of the NFT mania. After graduation, he joined crypto VC Distributed Global, spending his days “researching every area of crypto you can think of.” That habit of obsessive curiosity is one of the few traits he thinks good founders and good investors truly share. The turning point came when he met his co-founder Mike, who was being incubated by Polychain. @polychain and @SuiNetwork later backed Talus Network, and Ben crossed the aisle from investor to operator. As a founder, he says, “you need to be really, really optimistic, almost naively, foolishly optimistic,” whereas a VC’s job is to imagine every way something could fail. Living both roles helps him build ambitiously, but with a realistic sense of risk. Is There an AI Bubble? When asked if we are in an AI bubble, Ben said, “I don’t think that there’s an AI bubble. I think that multiples could come down, but a bubble would imply that the investment is a waste. I don’t think that it’s a waste relative to the opportunity that we have.” The real issue, he says, is that AI does not fit classic venture economics. Training competitive models means enormous hardware spend on day one, which is the opposite of the “two people in a garage” story VC was designed to fund. That is why much of the progress is happening inside large tech companies with deep balance sheets. For Ben, the right framing trumps financials. “The amount of good that this can do for the world is unlike anything we’ve ever seen.” If we get it right, AI changes how we work, learn, make decisions and even relate to each other. But that upside only matters if we tackle two big risks head-on. Big Risks: Wrong Models and Not Enough Power The first risk is model design. Today’s large language models are impressive but narrow: great at working over text, weak at time-series reasoning and embodied tasks like trading or robotics. Ben believes we will eventually need “world models” that learn more like humans if we want agents that can handle real-world complexity. If we stay locked into the current LLM paradigm, the flashy demos may plateau long before we reach the use cases people are dreaming about. The second risk is energy. Training and running frontier models is brutally power-hungry, and grids are already being stressed where data centres cluster. Ben is blunt here: the United States needs a “nuclear New Deal.” Modern small modular reactors are safer and cheaper, but the fuel, supply chain and expertise around them have withered. Rebuilding that, he argues, should be a top domestic priority: “There is no option other than nuclear energy” if America is serious about realising AI’s full potential. Why AI and Blockchain Actually Fit It is easy to dismiss “AI + crypto” as marketing, but Ben breaks the intersection into three clear layers: compute, data and agents. On the compute side, AI workloads need specialised hardware. Training is centralised and heavy; inference is lighter and can be spread out. Decentralised networks can pay people to contribute idle GPUs and other resources, especially for inference. In theory, your unused hardware could earn a small income answering model queries in the background. On the data side, the internet’s original sin was giving away our information for free. Now we know that “data is food for AI”, and better data makes stronger models. Blockchains allow people to own and permission their data, define who can access it, get paid for licensing and prove where it came from. Ben notes there are already many projects building that stack, and Talus already partners with several. The third layer is the one Talus is built for: agents. “An agent is just AI that is able to do something for you, like buy something for you,” he explains. In a world where you tell software, “rebalance my LPs” or “find me the best yield and move my money,” those agents will not be opening bank accounts, they will be using crypto. Talus’s Nexus framework puts an agent’s workflow on-chain while keeping inference off-chain. The result is something like an “AI smart contract” that is transparent, verifiable, composable with other programs and censorship-resistant. You cannot be banned from using an agent you depend on because a central provider changes its terms. That matters because, as Ben puts it, “the biggest thing that we need to solve for when it comes to AI is trust.” Deterministic Chains, Messy Reality Under the hood, Talus is wrestling with a classic blockchain constraint: determinism. On-chain systems need every state transition to be replayable, which is why DeFi uses oracles to bring in off-chain prices. Talus generalises that pattern to AI. A decentralised orchestrator layer aggregates model outputs generated off-chain and feeds them into on-chain workflows. That lets builders keep logic on deterministic infrastructure while still reacting to a messy, random world. DeFi is the first obvious playground: agents can auto-rebalance LP positions, move capital between strategies and handle the tedious parts of on-chain finance. Longer term, Ben is especially bullish on AI for entertainment. In an AI-driven world, he argues, “the last scarce resource on Earth will be human attention,” and agents that can attract and hold that attention, then plug it into prediction markets, tokens and games, will be hugely valuable. Why Sui, and Why Optimism Has to Be Earned Talus chose Sui as its base chain because of its Move language, object model and focus on transaction priority, which are all well-suited to a future where agents constantly send small, non-contentious payments to each other and to humans. Sui can stream those transactions efficiently while charging more when priority really matters. But Ben’s main takeaway is not about any specific chain. It is about attitude. “Be optimistic. Be really optimistic about AI. Be really optimistic about crypto. Be really optimistic about energy in the United States,” he says. Then comes the qualifier that defines his worldview: “Being optimistic is active. It’s not passive. You have to earn the right to be optimistic… you’ve got to work really hard to create the future that you envision.” Pessimism and cynicism may sound smarter, but Ben’s version of optimism is not wishful thinking. It is a call to build the energy systems, trust layers and crypto-native agents that could make that optimistic future real. 👉If you enjoyed reading the summary, head over to When Shift Happens on YouTube or your favorite podcast platform to access the full convo.
9
2
1
24
4,153
Not the beginning of the end, the end of the beginning
30
5
33
1,038
Overload the front of the curve and don’t be a doomer
I always love this stupid take It happens every year Because every year more us debt is being finance with t bills, so that looming number always looks imminent and scary T bills by definition are maturity less than 1 year. Therefore, all those bills are now issued at actually probably higher rates than right now since we’ve been in a cutting cycle So it’s actually probably a benefit to roll these bills at lower rates than what they were issued Nobody talks about the “problem” because it isn’t a problem
1
5
734